1 /*****************************************************************************
2 * subsdec.c : text subtitles decoder
3 *****************************************************************************
4 * Copyright (C) 2000-2006 the VideoLAN team
7 * Authors: Gildas Bazin <gbazin@videolan.org>
8 * Samuel Hocevar <sam@zoy.org>
9 * Derk-Jan Hartman <hartman at videolan dot org>
10 * Bernie Purcell <bitmap@videolan.org>
12 * This program is free software; you can redistribute it and/or modify
13 * it under the terms of the GNU General Public License as published by
14 * the Free Software Foundation; either version 2 of the License, or
15 * (at your option) any later version.
17 * This program is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 * GNU General Public License for more details.
22 * You should have received a copy of the GNU General Public License
23 * along with this program; if not, write to the Free Software
24 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
25 *****************************************************************************/
27 /*****************************************************************************
29 *****************************************************************************/
33 /*****************************************************************************
35 *****************************************************************************/
36 static int OpenDecoder ( vlc_object_t * );
37 static void CloseDecoder ( vlc_object_t * );
39 static subpicture_t *DecodeBlock ( decoder_t *, block_t ** );
40 static subpicture_t *ParseText ( decoder_t *, block_t * );
41 static char *StripTags ( char * );
42 static char *CreateHtmlSubtitle ( char * );
45 /*****************************************************************************
47 *****************************************************************************/
48 static const char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
49 "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
50 "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
51 "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
52 "ISO-8859-6", "CP1256", "MacArabic", "",
53 "ISO-8859-7", "CP1253", "MacGreek", "",
54 "ISO-8859-8", "CP1255", "MacHebrew", "",
55 "ISO-8859-9", "CP1254", "MacTurkish", "",
56 "ISO-8859-13", "CP1257", "",
57 "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
58 "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
59 "ISO-2022-KR", "EUC-KR", "",
60 "MacThai", "KOI8-T", "",
61 "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
62 "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
64 "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
65 "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
66 "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
67 "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
68 "HPROMAN8", "NEXTSTEP" };
70 SSA supports charset selection.
71 The following known charsets are used:
73 0 = Ansi - Western European
78 128 = Japanese (Shift JIS)
81 134 = GB2312 Simplified Chinese
82 136 = Big5 Traditional Chinese
89 204 = Russian (Cyrillic)
91 238 = Eastern European
95 static int pi_justification[] = { 0, 1, 2 };
96 static const char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
98 #define ENCODING_TEXT N_("Subtitles text encoding")
99 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
100 #define ALIGN_TEXT N_("Subtitles justification")
101 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
102 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
103 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
104 "UTF-8 encoding within subtitles files.")
105 #define FORMAT_TEXT N_("Formatted Subtitles")
106 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
107 "VLC partly implements this, but you can choose to disable all formatting.")
111 set_shortname( _("Subtitles"));
112 set_description( _("Text subtitles decoder") );
113 set_capability( "decoder", 50 );
114 set_callbacks( OpenDecoder, CloseDecoder );
115 set_category( CAT_INPUT );
116 set_subcategory( SUBCAT_INPUT_SCODEC );
118 add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
120 change_integer_list( pi_justification, ppsz_justification_text, 0 );
121 add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
122 ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
123 change_string_list( ppsz_encodings, 0, 0 );
124 add_bool( "subsdec-autodetect-utf8", VLC_TRUE, NULL,
125 AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, VLC_FALSE );
126 add_bool( "subsdec-formatted", VLC_TRUE, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
130 /*****************************************************************************
131 * OpenDecoder: probe the decoder and return score
132 *****************************************************************************
133 * Tries to launch a decoder and return score so that the interface is able
135 *****************************************************************************/
136 static int OpenDecoder( vlc_object_t *p_this )
138 decoder_t *p_dec = (decoder_t*)p_this;
139 decoder_sys_t *p_sys;
142 if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','u','b','t') &&
143 p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
148 p_dec->pf_decode_sub = DecodeBlock;
150 /* Allocate the memory needed to store the decoder's structure */
151 if( ( p_dec->p_sys = p_sys =
152 (decoder_sys_t *)calloc(1, sizeof(decoder_sys_t)) ) == NULL )
154 msg_Err( p_dec, "out of memory" );
160 p_sys->iconv_handle = (vlc_iconv_t)-1;
161 p_sys->b_autodetect_utf8 = VLC_FALSE;
162 p_sys->b_ass = VLC_FALSE;
163 p_sys->i_original_height = -1;
164 p_sys->i_original_width = -1;
165 TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
166 TAB_INIT( p_sys->i_images, p_sys->pp_images );
168 char *psz_charset = NULL;
169 /* First try demux-specified encoding */
170 if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
172 psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
173 msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
174 p_dec->fmt_in.subs.psz_encoding ?: "not specified");
177 /* Second, try configured encoding */
178 if (psz_charset == NULL)
180 psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
181 if ((psz_charset != NULL) && !strcasecmp (psz_charset, DEFAULT_NAME))
187 msg_Dbg (p_dec, "trying configured character encoding: %s",
188 psz_charset ?: "not specified");
191 /* Third, try "local" encoding with optional UTF-8 autodetection */
192 if (psz_charset == NULL)
194 psz_charset = strdup (GetFallbackEncoding ());
195 msg_Dbg (p_dec, "trying default character encoding: %s",
196 psz_charset ?: "not specified");
198 if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
200 msg_Dbg (p_dec, "using automatic UTF-8 detection");
201 p_sys->b_autodetect_utf8 = VLC_TRUE;
205 if (psz_charset == NULL)
207 psz_charset = strdup ("UTF-8");
208 msg_Dbg (p_dec, "trying hard-coded character encoding: %s",
209 psz_charset ?: "error");
212 if (psz_charset == NULL)
218 if (strcasecmp (psz_charset, "UTF-8") && strcasecmp (psz_charset, "utf8"))
220 p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
221 if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
222 msg_Err (p_dec, "cannot convert from %s: %s", psz_charset,
227 var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
228 var_Get( p_dec, "subsdec-align", &val );
229 p_sys->i_align = val.i_int;
231 if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
233 if( p_dec->fmt_in.i_extra > 0 )
234 ParseSSAHeader( p_dec );
240 /****************************************************************************
241 * DecodeBlock: the whole thing
242 ****************************************************************************
243 * This function must be fed with complete subtitles units.
244 ****************************************************************************/
245 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
247 subpicture_t *p_spu = NULL;
249 if( !pp_block || *pp_block == NULL ) return NULL;
251 p_spu = ParseText( p_dec, *pp_block );
253 block_Release( *pp_block );
259 /*****************************************************************************
260 * CloseDecoder: clean up the decoder
261 *****************************************************************************/
262 static void CloseDecoder( vlc_object_t *p_this )
264 decoder_t *p_dec = (decoder_t *)p_this;
265 decoder_sys_t *p_sys = p_dec->p_sys;
267 if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
268 vlc_iconv_close( p_sys->iconv_handle );
270 if( p_sys->pp_ssa_styles )
273 for( i = 0; i < p_sys->i_ssa_styles; i++ )
275 if( !p_sys->pp_ssa_styles[i] )
278 if( p_sys->pp_ssa_styles[i]->psz_stylename )
279 free( p_sys->pp_ssa_styles[i]->psz_stylename );
280 if( p_sys->pp_ssa_styles[i]->font_style.psz_fontname )
281 free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
282 if( p_sys->pp_ssa_styles[i] )
283 free( p_sys->pp_ssa_styles[i] );
285 TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
287 if( p_sys->pp_images )
290 for( i = 0; i < p_sys->i_images; i++ )
292 if( !p_sys->pp_images[i] )
295 if( p_sys->pp_images[i]->p_pic )
296 p_sys->pp_images[i]->p_pic->pf_release( p_sys->pp_images[i]->p_pic );
297 if( p_sys->pp_images[i]->psz_filename )
298 free( p_sys->pp_images[i]->psz_filename );
300 free( p_sys->pp_images[i] );
302 TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
308 /*****************************************************************************
309 * ParseText: parse an text subtitle packet and send it to the video output
310 *****************************************************************************/
311 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
313 decoder_sys_t *p_sys = p_dec->p_sys;
314 subpicture_t *p_spu = NULL;
315 char *psz_subtitle = NULL;
318 /* We cannot display a subpicture with no date */
319 if( p_block->i_pts == 0 )
321 msg_Warn( p_dec, "subtitle without a date" );
325 /* Check validity of packet data */
326 /* An "empty" line containing only \0 can be used to force
327 and ephemer picture from the screen */
328 if( p_block->i_buffer < 1 )
330 msg_Warn( p_dec, "no subtitle data" );
334 /* Should be resiliant against bad subtitles */
335 psz_subtitle = strndup( (const char *)p_block->p_buffer,
337 if( psz_subtitle == NULL )
340 if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
342 if (EnsureUTF8( psz_subtitle ) == NULL)
344 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
345 "Try manually setting a character-encoding "
346 "before you open the file.") );
352 if( p_sys->b_autodetect_utf8 )
354 if( IsUTF8( psz_subtitle ) == NULL )
356 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
357 "disabling UTF-8 subtitles autodetection" );
358 p_sys->b_autodetect_utf8 = VLC_FALSE;
362 if( !p_sys->b_autodetect_utf8 )
364 size_t inbytes_left = strlen( psz_subtitle );
365 size_t outbytes_left = 6 * inbytes_left;
366 char *psz_new_subtitle = malloc( outbytes_left + 1 );
367 char *psz_convert_buffer_out = psz_new_subtitle;
368 const char *psz_convert_buffer_in = psz_subtitle;
370 size_t ret = vlc_iconv( p_sys->iconv_handle,
371 &psz_convert_buffer_in, &inbytes_left,
372 &psz_convert_buffer_out, &outbytes_left );
374 *psz_convert_buffer_out++ = '\0';
375 free( psz_subtitle );
377 if( ( ret == (size_t)(-1) ) || inbytes_left )
379 free( psz_new_subtitle );
380 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
381 "Try manually setting a character-encoding "
382 "before you open the file.") );
386 psz_subtitle = realloc( psz_new_subtitle,
387 psz_convert_buffer_out - psz_new_subtitle );
391 /* Create the subpicture unit */
392 p_spu = p_dec->pf_spu_buffer_new( p_dec );
395 msg_Warn( p_dec, "can't get spu buffer" );
396 if( psz_subtitle ) free( psz_subtitle );
400 p_spu->b_pausable = VLC_TRUE;
402 /* Create a new subpicture region */
403 memset( &fmt, 0, sizeof(video_format_t) );
404 fmt.i_chroma = VLC_FOURCC('T','E','X','T');
406 fmt.i_width = fmt.i_height = 0;
407 fmt.i_x_offset = fmt.i_y_offset = 0;
408 p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
409 if( !p_spu->p_region )
411 msg_Err( p_dec, "cannot allocate SPU region" );
412 if( psz_subtitle ) free( psz_subtitle );
413 p_dec->pf_spu_buffer_del( p_dec, p_spu );
417 /* Decode and format the subpicture unit */
418 if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
420 /* Normal text subs, easy markup */
421 p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
422 p_spu->i_x = p_sys->i_align ? 20 : 0;
425 /* Remove formatting from string */
427 p_spu->p_region->psz_text = StripTags( psz_subtitle );
428 if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
430 p_spu->p_region->psz_html = CreateHtmlSubtitle( psz_subtitle );
433 p_spu->i_start = p_block->i_pts;
434 p_spu->i_stop = p_block->i_pts + p_block->i_length;
435 p_spu->b_ephemer = (p_block->i_length == 0);
436 p_spu->b_absolute = VLC_FALSE;
440 /* Decode SSA/USF strings */
441 if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
442 ParseSSAString( p_dec, psz_subtitle, p_spu );
444 p_spu->i_start = p_block->i_pts;
445 p_spu->i_stop = p_block->i_pts + p_block->i_length;
446 p_spu->b_ephemer = (p_block->i_length == 0);
447 p_spu->b_absolute = VLC_FALSE;
448 p_spu->i_original_picture_width = p_sys->i_original_width;
449 p_spu->i_original_picture_height = p_sys->i_original_height;
451 if( psz_subtitle ) free( psz_subtitle );
456 char* GotoNextLine( char *psz_text )
458 char *p_newline = psz_text;
460 while( p_newline[0] != '\0' )
462 if( p_newline[0] == '\n' || p_newline[0] == '\r' )
465 while( p_newline[0] == '\n' || p_newline[0] == '\r' )
474 /* Function now handles tags which has attribute values, and tries
475 * to deal with &' commands too. It no longer modifies the string
476 * in place, so that the original text can be reused
478 static char *StripTags( char *psz_subtitle )
480 char *psz_text_start;
483 psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
484 if( !psz_text_start )
487 while( *psz_subtitle )
489 if( *psz_subtitle == '<' )
491 if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
494 psz_subtitle += strcspn( psz_subtitle, ">" );
496 else if( *psz_subtitle == '&' )
498 if( !strncasecmp( psz_subtitle, "<", 4 ))
501 psz_subtitle += strcspn( psz_subtitle, ";" );
503 else if( !strncasecmp( psz_subtitle, ">", 4 ))
506 psz_subtitle += strcspn( psz_subtitle, ";" );
508 else if( !strncasecmp( psz_subtitle, "&", 5 ))
511 psz_subtitle += strcspn( psz_subtitle, ";" );
513 else if( !strncasecmp( psz_subtitle, """, 6 ))
516 psz_subtitle += strcspn( psz_subtitle, ";" );
520 /* Assume it is just a normal ampersand */
526 *psz_text++ = *psz_subtitle;
532 psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
534 return psz_text_start;
537 /* Try to respect any style tags present in the subtitle string. The main
538 * problem here is a lack of adequate specs for the subtitle formats.
539 * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
540 * SAMI has a detailed spec, but extensive rework is needed in the demux
541 * code to prevent all this style information being excised, as it presently
543 * That leaves the others - none of which were (I guess) originally intended
544 * to be carrying style information. Over time people have used them that way.
545 * In the absence of specifications from which to work, the tags supported
546 * have been restricted to the simple set permitted by the USF DTD, ie. :
547 * Basic: <br>, <i>, <b>, <u>
559 * There is also the further restriction that the subtitle be well-formed
560 * as an XML entity, ie. the HTML sentence:
561 * <b><i>Bold and Italics</b></i>
562 * doesn't qualify because the tags aren't nested one inside the other.
563 * <text> tags are automatically added to the output to ensure
565 * If the text doesn't qualify for any reason, a NULL string is
566 * returned, and the rendering engine will fall back to the
567 * plain text version of the subtitle.
569 static char *CreateHtmlSubtitle( char *psz_subtitle )
571 char psz_tagStack[ 100 ];
572 size_t i_buf_size = strlen( psz_subtitle ) + 100;
573 char *psz_html_start = malloc( i_buf_size );
575 psz_tagStack[ 0 ] = '\0';
577 if( psz_html_start != NULL )
579 char *psz_html = psz_html_start;
581 strcpy( psz_html, "<text>" );
584 while( *psz_subtitle )
586 if( *psz_subtitle == '\n' )
588 strcpy( psz_html, "<br/>" );
592 else if( *psz_subtitle == '<' )
594 if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
596 strcpy( psz_html, "<br/>" );
600 else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
602 strcpy( psz_html, "<b>" );
603 strcat( psz_tagStack, "b" );
607 else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
609 strcpy( psz_html, "<i>" );
610 strcat( psz_tagStack, "i" );
614 else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
616 strcpy( psz_html, "<u>" );
617 strcat( psz_tagStack, "u" );
621 else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
623 const char *psz_attribs[] = { "face=\"", "family=\"", "size=\"",
624 "color=\"", "outline-color=\"", "shadow-color=\"",
625 "outline-level=\"", "shadow-level=\"", "back-color=\"",
628 strcpy( psz_html, "<font " );
629 strcat( psz_tagStack, "f" );
633 while( *psz_subtitle != '>' )
637 for( k=0; psz_attribs[ k ]; k++ )
639 int i_len = strlen( psz_attribs[ k ] );
641 if( !strncasecmp( psz_subtitle, psz_attribs[ k ], i_len ))
643 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
645 strncpy( psz_html, psz_subtitle, i_len );
647 psz_subtitle += i_len;
651 if( psz_attribs[ k ] == NULL )
653 /* Jump over unrecognised tag */
654 int i_len = strcspn( psz_subtitle, "\"" ) + 1;
656 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
657 psz_subtitle += i_len;
659 while (*psz_subtitle == ' ')
660 *psz_html++ = *psz_subtitle++;
662 *psz_html++ = *psz_subtitle++;
664 else if( !strncmp( psz_subtitle, "</", 2 ))
666 vlc_bool_t b_match = VLC_FALSE;
667 int i_len = strlen( psz_tagStack ) - 1;
668 char *psz_lastTag = NULL;
672 psz_lastTag = psz_tagStack + i_len;
675 switch( *psz_lastTag )
678 b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
682 b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
686 b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
690 b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
697 /* Not well formed -- kill everything */
698 free( psz_html_start );
699 psz_html_start = NULL;
703 strncpy( psz_html, psz_subtitle, i_len );
705 psz_subtitle += i_len;
709 psz_subtitle += strcspn( psz_subtitle, ">" );
712 else if( *psz_subtitle == '&' )
714 if( !strncasecmp( psz_subtitle, "<", 4 ))
716 strcpy( psz_html, "<" );
720 else if( !strncasecmp( psz_subtitle, ">", 4 ))
722 strcpy( psz_html, ">" );
726 else if( !strncasecmp( psz_subtitle, "&", 5 ))
728 strcpy( psz_html, "&" );
734 strcpy( psz_html, "&" );
741 *psz_html = *psz_subtitle;
742 if( psz_html > psz_html_start )
744 /* Check for double whitespace */
745 if((( *psz_html == ' ' ) ||
746 ( *psz_html == '\t' )) &&
747 (( *(psz_html-1) == ' ' ) ||
748 ( *(psz_html-1) == '\t' )))
750 strcpy( psz_html, NO_BREAKING_SPACE );
751 psz_html += strlen( NO_BREAKING_SPACE ) - 1;
758 if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 10 )
760 int i_len = psz_html - psz_html_start;
763 psz_html_start = realloc( psz_html_start, i_buf_size );
764 psz_html = psz_html_start + i_len;
768 strcpy( psz_html, "</text>" );
771 if( psz_tagStack[ 0 ] != '\0' )
773 /* Not well formed -- kill everything */
774 free( psz_html_start );
775 psz_html_start = NULL;
777 else if( psz_html_start )
779 /* Shrink the memory requirements */
780 psz_html_start = realloc( psz_html_start, psz_html - psz_html_start + 1 );
783 return psz_html_start;